Social preferences in the public goods game-An Agent-Based simulation with EconSim

Area of Science:

• Behavioral Economics
• Computational Social Science
• Artificial Intelligence

Background:

• Public goods games are fundamental to understanding cooperation and social dilemmas.
• Endogenous punishment institutions play a crucial role in maintaining cooperation.
• Modeling human behavior in economic games requires sophisticated computational approaches.

Purpose of the Study:

• To develop an agent-based model simulating a public goods game with endogenous punishment.
• To propose and implement an outcome-based model of social preferences.
• To replicate and explain human behavior observed in experimental settings.

Main Methods:

• Utilized a reinforcement-learning algorithm for agent-based simulation.
• Developed an outcome-based utility function for agent decision-making.
• Compared simulation results with existing experimental data on public goods games.

Main Results:

• The reinforcement-learning model successfully replicated human contribution and voting behavior.
• The model explained the underlying motives driving agent decisions in the public goods game.
• Validated the outcome-based social preference model against experimental evidence.

Conclusions:

• Agent-based modeling with reinforcement learning can accurately simulate human behavior in social dilemmas.
• The proposed outcome-based social preference model provides insights into cooperation and punishment dynamics.
• The methodology offers a framework for generalizing to more complex simulations of human behavior.